Mind Leaps

After two philosophy-heavy posts, I’m taking a lighter turn this week to talk about something more immediately practical—at least, I hope so. When we talk about AI adoption in pharma and biotech, what are we really thinking—or more importantly, what should we really be thinking? Are we just chasing hype, or is AI truly going to change the way we work and what we produce?

To get there, I think we need to step back and build a general framework—not by going tech-first, but by starting with how AI actually functions today. That clarity will naturally guide business decisions. LLMs, at their core, are trained on a massive amount of human-generated data. They are exceptional at redistributing knowledge. So, I argue, when it comes to company-wide adoption of AI, viewing AI as a knowledge contributor—essentially, a teacher–helps us evaluate the strategic directions with a focused and clear lens. By “knowledge,” I mean structured, context-rich information that can be systematically applied to decision-making, workflows, and discovery—not raw data or isolated facts.

In parallel, the ultimate goal of adopting AI isn’t to use it piecemeal, or to dump it into every process possible. It is to build an AI-enabled company. Under this framework, the key strategic question becomes: What kind of knowledge actually makes an organization AI-enabled?

Using this framework, we can evaluate three layers of AI adoption, each tied to a different type of knowledge, with increasing level of importance to business success:

1. Cultural Knowledge (Transparency, Policy, and Values) – Easy Win

Undoubtedly, AI can transform how we work, especially when it comes to company culture. It can help disseminate institutional knowledge in ways that promote transparency and accessibility. In practice, this includes policies, internal processes, HR resources, career development pathways, and more. 

Leading companies are already implementing this approach. Salesforce, as an example, has implemented an AI-powered internal talent marketplace that provides transparency to career development. Rather than leaving institutional knowledge buried in outdated web portals that discourage access, or relying on word of mouth that depends on serendipity, AI helps create a culture of equal opportunity—powered by the free flow of information. This is a foundational step in increasing transparency and trust across the organization.

2. Technical/Operational Knowledge (Acceleration and Efficiency) – Harder Win

This domain is about using AI as an operational enabler to improve efficiency in areas like clinical trial protocol drafting, software development, and foundational infrastructure. Technical knowledge adoption is essential because it allows organizations to accelerate core business processes.

For instance, Novo Nordisk has adopted Claude to streamline the technical writing process for compliance documents. However, when considering institutional adoption, without strong cultural and strategic alignment, AI often gets deployed into siloed teams—leading to duplicated efforts, wasted resources, and projects that exist more for optics than real impact. It is therefore critical to ground technical adoption in shared priorities and real, non-performative, cross-functional collaboration.

3. Scientific/Innovative Knowledge (Strategic Differentiation) – Hardest Win

If you ask me what is most transformative about becoming AI-enabled, I’ll answer without hesitation: AI enables innovation. It pushes our core competencies forward. Scientific knowledge is the hardest, and the most potentially transformative, domain for AI adoption. 

It includes:

• Hypothesis generation and integration through cross-domain connections and novel pattern recognition
• Strategic scientific focus driven by clearly defined, high-priority scientific questions

Scientific research is already prone to bias, cherry-picked findings, and flexible interpretation. AI cannot correct for that on its own—in fact, it can amplify it (see my post Group-thinking: What it Means for AI Safety for a deep dive). That’s why human expertise is essential, even when evaluating widely accepted beliefs. To truly succeed in AI-enabled innovation, companies need the best analytical minds—not the political ones—who can interrogate, refine, and meaningfully collaborate with AI. No question: AI-enabled means talent-enabled.

Final Thoughts

An AI-enabled company is not defined by how much AI it uses. It is defined by how strategically it integrates AI-driven knowledge into its cultural DNA, operational processes, and scientific/innovative direction. It surely is not about hype, but about deliberately shaping the organization’s relationship with knowledge. The future belongs to organizations that treat AI as a core strategic asset—driving transparency, trust, empowerment, alignment, and innovation from the inside out.